The Model Artificial Intelligence Law (MAIL) v.4.0

The Model Artificial Intelligence Law 4.0 (“MAIL 4.0”) offers a comprehensive statutory blueprint for governing artificial intelligence while preserving the institutional conditions for its development. Its central premise is that innovation and regulation are not competing objectives, but mutually dependent components of a durable AI governance regime. To that end, the Model Law combines extensive public support for computing infrastructure, data resources, basic research, open-source development, and industrial experimentation with a differentiated system of legal oversight. At the center of that system is a dynamically updated negative list: AI activities presenting heightened risks are subject to ex ante testing, while other providers may be governed through registries and continuing oversight. MAIL 4.0 allocates responsibility across the AI lifecycle, distinguishing among developers, providers, and users and imposing role-specific duties concerning safety and security, transparency, explainability, fairness, auditing, vulnerability management, human oversight, content labeling, and emergency intervention. It supplements these duties with specialized rules for foundation and derivative models, generative AI, open-source systems, AI agents, employment, scientific research, and governmental decision-making. The proposal also addresses the institutional dynamics of AI through copyright licensing rules, innovation zones, regulatory sandboxes, public computing resources, tax incentives, whistleblower protections, and calibrated safe harbors. Institutionally, the Model Law would establish a national AI administrative authority with powers extending from standard-setting and risk monitoring to licensing, enforcement, and extraterritorial regulation. Civil, administrative, and public-interest remedies complete the framework. MAIL 4.0’s distinctive contribution lies less in any single regulatory instrument than in its attempt to integrate developmental policy, lifecycle governance, public-law accountability, and private-law liability within a unified legislative architecture. It therefore provides both a China-oriented model of AI legislation and a broader framework for comparative debate over how law should govern rapidly evolving, general-purpose technologies without freezing their development.